MODEL ȘI ANALOGIE: RAPORTURI EPISTEMICE ISTORICE

Abstract

The basis of the article is a semantic research of the concept of model: and its comparison with those of "concept", "theory", "paradigm" allows us to see what the specificity of the model as a cognitive approach consists of. However, abstract concepts – like those above – are ambiguous, and a quick point of their ambiguity helps us perceive both the problem of ambiguity itself and the concrete problem of distinguishing the "model". An essential means of constructing models is analogy. Highlighting the different aspects of its definition allows us to capture the evolution of scientific models according to the changing types of analogy. The model of science here is physics. Traditionally, the scientific model was based on analogies with natural phenomena. Because mathematics uses internal analogies, between mathematical problems and solutions, the mathematization of physics has led to the drastic limitation of analogies with natural phenomena and to the cascading emergence of models of physics. With all the physical object of physics and with all the "application" of mathematics to physical reality, mathematization has led to a production of models from models, hence from the reality of abstract mathematical objects, from the reality of abstract creation.
But Artificial Intelligence (AI) – which is a production and offer of models for solving real problems in knowledge and the real world – is, although internally constituted on mathematical models, a creation through analogies with the real physical world. AI is trained/loaded with data and data processing algorithms. In principle and in perspective, the data loaded into AI is much more than that in the mind of a researcher. As a result, the analogies are also more numerous and more original, and, thus, the problem-solving models are better. And, just as the mathematization of sciences led to the forgetting of the initial analogies and the emergence of formalized models, so too AI creatively develops models from other models whose factual origin remained in the history of the cognitive approach.